Dynamic Route Failover

[Jesse Hottle]

Hello,

I am trying to get our network traffic to dynamically pick the best route out/in in case of a link failure. I have attached a diagram showing our current router links. The majority of the links are connected to the end-point routers using wireless backhaul bridges. We have two connections forming a ring back to our central noc. (DS3, and PTP fiber connections). Can any one help with this?

[Jesse Hottle]

I am going to put together a new diagram, showing the non-backbone attached routers as well as configs for some of the routers. I also think the way we have some of our private IP ranges defined, is causing some issues and preventing me from making some of the changes needed.

I'll post ASAP.

Thanks,,

j

[Jesse Hottle]

Hello,

I have attached a full network diagram. This is accurate. I have also attached dumps for some of the connected routers. The interface labels on the diagram to no match the dumps, not to confuse you.

On the diagram router M is the single network we discussed adding to its own area, which I want to do. Little confused on the setup, but I dont think it will be an issue. Our biggest problem is that we sigle class "C" subnets spilt up across multiple routers. I dont know how that would be summarized. Each router has multiple hosts and devices attached. I did not put that in the diagram.

j

[Jesse Hottle]

remaining posts... more to follow. 3 attachment limit.

[Jesse Hottle]

final attachments

[Istvan_Rabai]

Hi Jesse,

I have just downloaded the files. I will look over the configs and return to you.

This may take some of out of work time, so it can happen that I will return to you tomorrow or after tomorrow.

Cheers:

Istvan

[Jesse Hottle]

Cool and thanks. I am going to start resubnetting parts of the network off the 192.168.x.x subnets and take your advice on using a 10.x.x.x approach. Think this will greatly clean up the routing tables and be better organized.

thanks again,.

[Jesse Hottle]

Got a quick question.

One on of the 3550 in the backbone network, I have a routed vlan setup. I added the following address range to this vlan: 10.1.0.0/20

Of this network I have 9 active subnets:

10.1.0.0/24

10.1.1.0/24

10.1.2.0/24

10.1.3.0/24

10.1.4.0/24

10.1.5.0/24

10.1.6.0/24

10.1.7.0/24

10.1.8.0/24

10.1.9.0/24

Is there a way to summarize this network range even though these are subnets on this 3550? Under the ospf network settings I added network 10.1.0.0 0.0.15.255 to be advertized, but the network range is not being summarized. Routing tables show each individual subnet. I am doing this to try and reduce the size of the routing tables.

j

[Istvan_Rabai]

Hi Jesse,

You need to summarize this address range under the ospf process like this:

router ospf 1

area x range 10.1.0.0 0.0.15.255

where x is the area number where these subnets are located.

This will advertise the summarized address into area 0.

Cheers:

Istvan

[Jesse Hottle]

problem I have is this is a l3 switch, as part of the backbone routing on area 0

Can I setup the routed vlans on this switch into different areas?

j

[Istvan_Rabai]

Yes, you can.

But, if you put all the vlans in different areas then you will not be able to summarize the addresses.

The area range command works on one area only, it doesn't work for multiple areas.

This is why I suggest to put the routed vlans in the same area.

Routing will happen between them anyway.

I quickly went through the config of Router C (3550) and at first glance I could see the following:

On interfaces vlan2, vlan3 and vlan5 you configured multiple secondary ip addresses.

This is not OK. Only one primary and one secondary address can be configured (as far as I know).

This probably causes routing problems for you.

Please remove the additional secondary ip addresses and leave only one primary and one secondary ip address.

Later I will review the configs more thoroughly. This just my first quick look.

Cheers:

Istvan

[Jesse Hottle]

Hello,

I cant only have two ip addresses per interface or vlan. I have many devices and customers behind each inteface or vlan where I need these IP's. I have not had any issues with this in the past. We are moving to installing routers at our remote access locations, which will help remove a lot of the secondary addresses, but not for some time.

j

[Jesse Hottle]

Hello,

Any word?

Thanks,

Jesse

[Istvan_Rabai]

Hi Jesse,

On Router A apply the "default-information originate metric-type 1" command.

This will add a cumulating cost to the default cost of 1 for the default route as it is propagating in the ospf domain.

This way the ospf processes of internal routers will have information on wich route is better to Router A when they want to route packets to the outside world.

They will put the default route with the least cost into their routing tables.

Please refer back if this solved the issue.

Cheers:

Istvan

[Istvan_Rabai]

Hi Jesse,

I had a chance to look over your configurations and here are my observations:

Router A:

For security reasons, it is necessary to disable cdp only towards the ISP (no cdp enable).

On other interfaces you can leave it enabled, if no security reasons exist. This helps you in troubleshooting the network whenever necessary.

As I mentioned, you use many secondary ip addresses, like on Router A:

interface Ethernet4/1/0

ip address x.x.80.33 255.255.255.248 secondary

ip address 192.168.30.1 255.255.255.0 secondary

ip address x.x.80.129 255.255.255.128 secondary

ip address 192.168.200.1 255.255.255.0 secondary

ip address x.x.158.1 255.255.255.252 secondary

ip address x.x.81.177 255.255.255.240

ip helper-address x.x.89.18

no ip redirects

no ip unreachables

half-duplex

no cdp enable

If you want to serve more than 2 subnets on the same router interface then I suggest configuring vlans for each subnet. Then you can configure routing for the vlans the following way:

interface Ethernet4/1/0

no ip address

interface Ethernet4/1/0.1 [subinterface for vlan 1]

ip address x.x.81.177 255.255.255.240

encapsulation dot1q 1 native

interface Ethernet4/1/0.2 [subinterface for vlan 2]

ip address x.x.80.33 255.255.255.248

encapsulation dot1q 2

interface Ethernet4/1/0.3 [subinterface for vlan 3]

ip address x.x.80.129 255.255.255.128

encapsulation dot1q 3

interface Ethernet4/1/0.4 [subinterface for vlan 4]

ip address x.x.158.1 255.255.255.252

encapsulation dot1q 4

interface Ethernet4/1/0.5 [subinterface for vlan 5]

ip address 192.168.30.1 255.255.255.0

encapsulation dot1q 5

interface Ethernet4/1/0.6 [subinterface for vlan 6]

ip address 192.168.200.1 255.255.255.0

encapsulation dot1q 6

ip helper-address x.x.89.18

half-duplex

As you configured interface Ethernet4/1/0 as half-duplex, I suppose you have a hub connected to this interface.

For the above to work, you need to replace the hub with a layer2 switch, like a 2950 or 2960. You need to configure the vlans within the switch and connect the switch to interface Ethernet4/1/0 (configured for full-duplex)

You also need to configure the layer2 switch interface as a trunk:

Example:

interface Fastethernet 1/0

speed 10

duplex full

switchport trunk encapsulation dot1q

switchport trunk native vlan 1

switchport trunk allowed vlan 1-6

switchport mode trunk

This way you will separate the different subnets from each other, but they will get their traffic routed to the outside world (and between each other).

OSPF:

Try to design a hierarchical addressing scheme.

On interface Ethernet4/1/0 alone you used 6 subnets, but they are not in the same supernet, so it is very difficult to summarize them and advertise them as 1 network. This is why you have so many routes in the routing tables.

You can put these subnets in an area different than area 0.

If these subnets would be summarizable, then you could advertise them into area 0 (the backbone) with the "area x range" command as I described earlier.

The same applies to all the other routers where you have many secondary ip addresses:

Router C (3550)

Put the different subnets in different vlans. The 3550 will route the traffic between them anyway. You just need to configure some more vlan interfaces.

Put these subnets again in an area different than area 0.

Also, by applying summarization, your routing tables will be greatly reduced.

In addition to less computation needed for the routing protocol, it will also ease the readability of the configuration and troubleshooting now and in the future.

Cheers:

Istvan

[Jesse Hottle]

Hello,

Problem I have with the vlans is that way we are feeding some of these remote tower sites and customers. There really are not seperate connections to vlan out.

Example. I have an access unit, feeding Internet customers and multiple tower locations, which are also serving Internet to customers. The AP serving these locations and people is connected to a single switch port on a 3550, on it's own routed vlan. So, I might have 400 customers, all getting routable IP address, but each tower unit, subscriber modem, all have their own private IP for management purposed. This is why we have so many secondary IP address per vlan / interface. We are working off of this setup, just gonna take time, which is lacking.

If I want to create a stub ospf network, can I summarize some of the IP addresses within the stub network? I ask this because I have different subnets from completely different subnets.

Thanks for all the assistance.

j